pandora-kernel.git
14 years agoCAPI: Remove experimental tag from middleware feature
Jan Kiszka [Mon, 8 Feb 2010 10:12:45 +0000 (10:12 +0000)]
CAPI: Remove experimental tag from middleware feature

Despite all its bugs, the middleware support of our CAPI stack was
already in use for many, many moons. And after going through its code,
fixing all issues I found, I feel it deserves to officially become a
non-experimental feature.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Schedule capifs for removal
Jan Kiszka [Mon, 8 Feb 2010 10:12:44 +0000 (10:12 +0000)]
CAPI: Schedule capifs for removal

With dynamic TTY nodes and the help of udev, we no longer need this
special filesystem. Schedule it for removal in one year from now.

As a last duty to this feature, move its help to right option so that
users can read the rationale.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Drop special controller lookup from capi20_put_message
Jan Kiszka [Mon, 8 Feb 2010 10:12:43 +0000 (10:12 +0000)]
CAPI: Drop special controller lookup from capi20_put_message

This strange special rule to fall back to controller 1 cannot be derived
from the CAPI specs and looks a lot like it was once dedicated to some
out-of-tree driver, probably AVM's broken fcdsl2 (FRITZ!Card DSL v2.0).
I found no in-tree user that needs this check, and I'm now taking care
of the fcdsl2. So drop these bits from our stack.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Drop return value of handle_minor_send
Jan Kiszka [Mon, 8 Feb 2010 10:12:42 +0000 (10:12 +0000)]
CAPI: Drop return value of handle_minor_send

We did not evaluate handle_minor_send's return value, just (void)'ed it
away. Time for a cleanup.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Clean up capiminor_*_ack
Jan Kiszka [Mon, 8 Feb 2010 10:12:41 +0000 (10:12 +0000)]
CAPI: Clean up capiminor_*_ack

No need for irqsave acquisition of acklock, bh-safe is sufficient.
Moverover, move kfree out of the lock and do not take acklock at all
in capiminor_del_all_ack as we are the last user of the list here.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Fix locking around capiminor's output queue and drop workaround_lock
Jan Kiszka [Mon, 8 Feb 2010 10:12:40 +0000 (10:12 +0000)]
CAPI: Fix locking around capiminor's output queue and drop workaround_lock

Introduce outlock as a spin lock that protects capiminor's outqueue,
outbytes and outskb (formerly known as ttyskb). outlock can be acquired
from soft-IRQ context via capinc_write, so make it bh-safe.

This finally removes the last reason for keeping the workaround lock
around (which was incomplete and partly broken anyway). And as we no
longer call handle_recv_skb in atomic context, gen_data_b3_resp_for can
use non-atomic allocation now.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Drop capiminor's unused inbytes counter
Jan Kiszka [Mon, 8 Feb 2010 10:12:39 +0000 (10:12 +0000)]
CAPI: Drop capiminor's unused inbytes counter

The inbytes counter was only updated but never read.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Use atomics for capiminor's datahandle and msgid
Jan Kiszka [Mon, 8 Feb 2010 10:12:38 +0000 (10:12 +0000)]
CAPI: Use atomics for capiminor's datahandle and msgid

The capiminor members datahandle and msgid are incremented outside any
lock, so better do this atomically.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Rename datahandle_queue -> ackqueue_entry
Jan Kiszka [Mon, 8 Feb 2010 10:12:37 +0000 (10:12 +0000)]
CAPI: Rename datahandle_queue -> ackqueue_entry

This struct is describing a queue entry, not the queue itself.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Rework capiminor RX handler
Jan Kiszka [Mon, 8 Feb 2010 10:12:36 +0000 (10:12 +0000)]
CAPI: Rework capiminor RX handler

Avoid re-queuing skbs unless the error detected in handle_recv_skb is
expected to be recoverable such as lacking memory, a full CAPI queue, a
full TTY input buffer, or a not yet existing TTY.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Drop handle_minor_recv from capinc_tty_write
Jan Kiszka [Mon, 8 Feb 2010 10:12:35 +0000 (10:12 +0000)]
CAPI: Drop handle_minor_recv from capinc_tty_write

Sending a message down the CAPI stack may trigger the reception of an
answer, but this will go through capi_recv_message and call
handle_minor_recv from there. There is no need to walk the receive queue
on capinc_tty_write.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Drop atomic ttyopencount
Jan Kiszka [Mon, 8 Feb 2010 10:12:34 +0000 (10:12 +0000)]
CAPI: Drop atomic ttyopencount

Not needed, tty->count keeps track of this information. At this chance,
drop traces of ancient attempts to debug this logic via _DEBUG_REFCOUNT.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Clean up capiminors_lock
Jan Kiszka [Mon, 8 Feb 2010 10:12:33 +0000 (10:12 +0000)]
CAPI: Clean up capiminors_lock

Use a plain spin lock for capiminors_lock, drop inconsistent irqsafe
acquisitions (it's only used in process context anyway).

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Drop obsolete nccip from capiminor struct
Jan Kiszka [Mon, 8 Feb 2010 10:12:32 +0000 (10:12 +0000)]
CAPI: Drop obsolete nccip from capiminor struct

The nccip in capiminor used to serve as an indicator that the NCCI was
close. But we don't need this, we issue a hangup on capincci_free_minor.
So drop this legacy.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Issue synchronous hangup on capincci_free_minor
Jan Kiszka [Mon, 8 Feb 2010 10:12:31 +0000 (10:12 +0000)]
CAPI: Issue synchronous hangup on capincci_free_minor

capincci_free and, thus, capincci_free_minor runs in process context, so
we can issue the hangup of the associated TTY synchronously.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Drop remaining NULL checks on tty->driver_data
Jan Kiszka [Mon, 8 Feb 2010 10:12:30 +0000 (10:12 +0000)]
CAPI: Drop remaining NULL checks on tty->driver_data

tty_struct's driver_data cannot be NULL, no need to test for it.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Use tty_port to keep track of capiminor's tty
Jan Kiszka [Mon, 8 Feb 2010 10:12:29 +0000 (10:12 +0000)]
CAPI: Use tty_port to keep track of capiminor's tty

Use the reference management features of tty_port to look up and drop
again the tty_struct associated with a capiminor.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Establish install/cleanup handlers for capiminor TTYs
Jan Kiszka [Mon, 8 Feb 2010 10:12:28 +0000 (10:12 +0000)]
CAPI: Establish install/cleanup handlers for capiminor TTYs

Properly associate/disassociate a capiminor object with its TTY via the
install/cleanup handlers instead of trying to guess first open and last
close.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Use kref on capiminor
Jan Kiszka [Mon, 8 Feb 2010 10:12:27 +0000 (10:12 +0000)]
CAPI: Use kref on capiminor

Install a reference counter for capiminor objects. Acquire it when
obtaining a capiminor from the array during capinc_tty_open, drop it
when closing the tty again. Another reference is held for the hook-up
with capincci.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Use dynamic major for NCCI TTYs by default
Jan Kiszka [Mon, 8 Feb 2010 10:12:26 +0000 (10:12 +0000)]
CAPI: Use dynamic major for NCCI TTYs by default

No need to allocate a fixed major for this TTY, both capifs and udev
make this transparent to the user.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Dynamically register minor devices
Jan Kiszka [Mon, 8 Feb 2010 10:12:25 +0000 (10:12 +0000)]
CAPI: Dynamically register minor devices

Register capiminors dynamically with the TTY core so that udev can make
them show up as the NCCIs appear or disappear. This removes the need to
check if the capiminor requested in capinc_tty_open actually exists.

And this completely obsoletes capifs which will be scheduled for removal
in a later patch.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Clean up capinc_tty_init/exit
Jan Kiszka [Mon, 8 Feb 2010 10:12:24 +0000 (10:12 +0000)]
CAPI: Clean up capinc_tty_init/exit

Return proper error code if tty_register_driver fails. In contrast,
tty_unregister_driver cannot practically fail, so drop that error
handling. Finally, mark capinc_tty_init/exit with __init/__exit.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Switch capiminor list to array
Jan Kiszka [Mon, 8 Feb 2010 10:12:23 +0000 (10:12 +0000)]
CAPI: Switch capiminor list to array

Using a plain array of pointers simplifies the management of capiminors.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Switch NCCI list to standard doubly linked list
Jan Kiszka [Mon, 8 Feb 2010 10:12:22 +0000 (10:12 +0000)]
CAPI: Switch NCCI list to standard doubly linked list

Replace open-coded NCCI list management with standard mechanisms.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Fix racy capi_read
Jan Kiszka [Mon, 8 Feb 2010 10:12:21 +0000 (10:12 +0000)]
CAPI: Fix racy capi_read

capi_read still used interruptible_sleep_on, risking to miss a wakeup
this way. Convert it to wait_event_interruptible.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Use non-atomic allocation during NCCI setup
Jan Kiszka [Mon, 8 Feb 2010 10:12:20 +0000 (10:12 +0000)]
CAPI: Use non-atomic allocation during NCCI setup

Both capincci_alloc and capiminor_alloc run in non-atomic context,
update their memory allocations accordingly.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Rework locking of capidev members
Jan Kiszka [Mon, 8 Feb 2010 10:12:19 +0000 (10:12 +0000)]
CAPI: Rework locking of capidev members

Rename 'ncci_list_mtx' to 'lock', expressing that it now protects a
larger set of capidev members: the NCCI list, ap.applid (ie. the
registration of the application), and modifications of userflags.

We do not need to protect each and every check for ap.applid because,
once an application is registered, it will stay for the whole lifetime
of the device.

Also, there is no need to apply the capidev mutex during release (if
there could be concurrent users, we would crash them anyway by freeing
the device at the end of capi_release).

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Clean up capi_open/release
Jan Kiszka [Mon, 8 Feb 2010 10:12:18 +0000 (10:12 +0000)]
CAPI: Clean up capi_open/release

Fold capidev_alloc and capidev_free into capi_open and capi_release -
there are no other users. Someone pushed a lock_kernel into capi_open.
Drop it, we don't need it. Also remove the useless test from open that
checks for private_data == NULL.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Convert capidev_list_lock into a mutex
Jan Kiszka [Mon, 8 Feb 2010 10:12:17 +0000 (10:12 +0000)]
CAPI: Convert capidev_list_lock into a mutex

No need for anything "harder" here (specifically no need for
irqsave...). Also, make the list removal the first operation of
capidev_free to avoid dumping half-released devices via /proc.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Reduce #ifdef mess around CONFIG_ISDN_CAPI_MIDDLEWARE
Jan Kiszka [Mon, 8 Feb 2010 10:12:16 +0000 (10:12 +0000)]
CAPI: Reduce #ifdef mess around CONFIG_ISDN_CAPI_MIDDLEWARE

Make the code a bit more readable be providing stub functions for the
!CONFIG_ISDN_CAPI_MIDDLEWARE case. Though a few lines are moved around,
this comes with no functional changes.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Rework application locking
Jan Kiszka [Mon, 8 Feb 2010 10:12:15 +0000 (10:12 +0000)]
CAPI: Rework application locking

Drop the application rw-lock in favour of RCU. This synchronizes
capi20_release against capi_ctr_handle_message which may dereference an
application from (soft-)IRQ context. Any other access to the application
list is now protected by the capi_controller_lock as well. This also
allows to safely inspect applications for /proc dumping by holding
capi_controller_lock.

At this chance, drop some useless release_in_progress checks where we
obtained the application pointer from the list (which becomes NULL on
release_in_progress).

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Rework locking of controller data structures
Jan Kiszka [Mon, 8 Feb 2010 10:12:14 +0000 (10:12 +0000)]
CAPI: Rework locking of controller data structures

This patch applies the mutex so far only protecting the controller list
to (almost) all accesses of controller data structures. It also reworks
waiting on state changes in old_capi_manufacturer so that it no longer
poll and holds a module reference to the controller owner while waiting
(the latter was partly done already). Modification and checking of the
blocked state remains racy by design, the caller is responsible for
dealing with this.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Rework controller state notifier
Jan Kiszka [Mon, 8 Feb 2010 10:12:13 +0000 (10:12 +0000)]
CAPI: Rework controller state notifier

Another step towards proper locking: Rework the callback provided to
capidrv for controller state changes. This is so far attached to an
application, which would require us to hold the corresponding lock
across notification calls.

But there is no direct relation between a controller up/down event and
an application, so let's decouple them and provide a notifier call chain
for those events instead. This notifier chain is first of all used
internally. Here we request the highest priority to unsure that
housekeeping work is done before any other notifications. The chain is
exported via [un]register_capictr_notifier to our only user, capidrv, to
replace the racy and unfixable capi20_set_callback.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Rework capi_ctr_ready/down
Jan Kiszka [Mon, 8 Feb 2010 10:12:12 +0000 (10:12 +0000)]
CAPI: Rework capi_ctr_ready/down

This step prepares the application of proper controller locking: Push
all state changing work into the notify handler that are called by
capi_ctr_ready and capi_ctr_down, switch detach_capi_ctr to issue a
synchronous ctr_down. Also ensure that we do not go through any action
if the state did not change.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Convert capi drivers rwlock into mutex
Jan Kiszka [Mon, 8 Feb 2010 10:12:11 +0000 (10:12 +0000)]
CAPI: Convert capi drivers rwlock into mutex

Turn the lock protecting registered capi drivers into a mutex and apply
it consistently.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Call a controller 'controller', not 'card'
Jan Kiszka [Mon, 8 Feb 2010 10:12:10 +0000 (10:12 +0000)]
CAPI: Call a controller 'controller', not 'card'

At least for our internal use, fix the misnomers that refer to a CAPI
controller as 'card'. No functional changes.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Reduce chattiness during module loading/removal
Jan Kiszka [Mon, 8 Feb 2010 10:12:09 +0000 (10:12 +0000)]
CAPI: Reduce chattiness during module loading/removal

The CVS revisions dumped by all CAPI modules are meaningless today. And
that some CAPI module is loaded or removed does not necessarily deserve
a message. Just keep the message of the central module, capi.ko, drop
the rest.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Pin capifs instead of mounting it
Jan Kiszka [Mon, 8 Feb 2010 10:12:08 +0000 (10:12 +0000)]
CAPI: Pin capifs instead of mounting it

Auto-mounting the capifs during module init prevents unloading its
module. Instead, pin the filesystem as long as some NCCI node exists.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Eliminate capifs_root variable
Jan Kiszka [Mon, 8 Feb 2010 10:12:07 +0000 (10:12 +0000)]
CAPI: Eliminate capifs_root variable

capifs_mnt->mnt_sb->s_root already contains what we need.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Sanitize capifs API
Jan Kiszka [Mon, 8 Feb 2010 10:12:06 +0000 (10:12 +0000)]
CAPI: Sanitize capifs API

Instead of looking up the dentry of an NCCI node again in
capifs_free_ncci pass the pointer via the capifs user.

This patch also reduces the #ifdef mess in capi.c a bit as far as capifs
was causing it.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoCAPI: Fix leaks in capifs_new_ncci
Jan Kiszka [Mon, 8 Feb 2010 10:12:05 +0000 (10:12 +0000)]
CAPI: Fix leaks in capifs_new_ncci

When something went wrong during capifs_new_ncci, the looked up dentry
was not properly released. Neither was the allocated inode. Refactor the
function to avoid leaks.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoserial: 8250: add serial transmitter fully empty test
Dick Hollenbeck [Wed, 9 Dec 2009 20:31:34 +0000 (12:31 -0800)]
serial: 8250: add serial transmitter fully empty test

When controlling an industrial radio modem it can be necessary to
manipulate the handshake lines in order to control the radio modem's
transmitter, from userspace.

The transmitter should not be turned off before all characters have been
transmitted.  serial8250_tx_empty() was reporting that all characters were
transmitted before they actually were.

===

Discovered in parallel with more testing and analysis by Kees Schoenmakers
as follows:

I ran into an NetMos 9835 serial pci board which behaves a little
different than the standard.  This type of expansion board is very common.

"Standard" 8250 compatible devices clear the 'UART_LST_TEMT" bit together
with the "UART_LSR_THRE" bit when writing data to the device.

The NetMos device does it slightly different

I believe that the TEMT bit is coupled to the shift register.  The problem
is that after writing data to the device and very quickly after that one
does call serial8250_tx_empty, it returns the wrong information.

My patch makes the test more robust (and solves the problem) and it does
not affect the already correct devices.

Alan:

  We may yet need to quirk this but now we know which chips we have a
  way to do that should we find this breaks some other 8250 clone with
  dodgy THRE.

Signed-off-by: Dick Hollenbeck <dick@softplc.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Cc: Kees Schoenmakers <k.schoenmakers@sigmae.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agonet neigh: Decouple per interface neighbour table controls from binary sysctls
Eric W. Biederman [Sun, 14 Feb 2010 03:27:03 +0000 (03:27 +0000)]
net neigh: Decouple per interface neighbour table controls from binary sysctls

Stop computing the number of neighbour table settings we have by
counting the number of binary sysctls.  This behaviour was silly
and meant that we could not add another neighbour table setting
without also adding another binary sysctl.

Don't pass the binary sysctl path for neighour table entries
into neigh_sysctl_register.  These parameters are no longer
used and so are just dead code.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet ipv4: Decouple ipv4 interface parameters from binary sysctl numbers
Eric W. Biederman [Sun, 14 Feb 2010 03:25:51 +0000 (03:25 +0000)]
net ipv4: Decouple ipv4 interface parameters from binary sysctl numbers

Stop using the binary sysctl enumeartion in sysctl.h as an index into
a per interface array.  This leads to unnecessary binary sysctl number
allocation, and a fragility in data structure and implementation
because of unnecessary coupling.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/pcnet32.c: Checkpatch cleaning
Joe Perches [Mon, 15 Feb 2010 08:34:26 +0000 (08:34 +0000)]
drivers/net/pcnet32.c: Checkpatch cleaning

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/pcnet32.c: Use (pr|netdev|netif)_<levels> macro helpers
Joe Perches [Mon, 15 Feb 2010 08:34:25 +0000 (08:34 +0000)]
drivers/net/pcnet32.c: Use (pr|netdev|netif)_<levels> macro helpers

Make the output logging messages a bit more consistent.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/tehuti.c: trivial checkpatch cleanups
Joe Perches [Mon, 15 Feb 2010 08:34:24 +0000 (08:34 +0000)]
drivers/net/tehuti.c: trivial checkpatch cleanups

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/tehuti.c: Hoist assigns out of ifs
Joe Perches [Mon, 15 Feb 2010 08:34:23 +0000 (08:34 +0000)]
drivers/net/tehuti.c: Hoist assigns out of ifs

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/tehuti.c: Use (pr|netdev|netif)_<levels> macro helpers
Joe Perches [Mon, 15 Feb 2010 08:34:22 +0000 (08:34 +0000)]
drivers/net/tehuti.c: Use (pr|netdev|netif)_<levels> macro helpers

Make the output logging messages a bit more consistent.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/sky2: Convert to use netif_printk macros
Joe Perches [Mon, 15 Feb 2010 08:34:21 +0000 (08:34 +0000)]
drivers/net/sky2: Convert to use netif_printk macros

Some unlikely(netif_msg_<foo>(sky2)) tests are also
removed by this change.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/ixgb: Use netif_printk macros
Joe Perches [Mon, 15 Feb 2010 08:34:20 +0000 (08:34 +0000)]
drivers/net/ixgb: Use netif_printk macros

Convert private DPRINTK macro uses to netif_<level> equivalents
Remove #define DPRINTK

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoclass: Free the class private data in class_release
Laurent Pinchart [Wed, 10 Feb 2010 12:32:49 +0000 (13:32 +0100)]
class: Free the class private data in class_release

Fix a memory leak by freeing the memory allocated in __class_register
for the class private data.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agosysfs: sysfs_sd_setattr set iattrs unconditionally
Eric W. Biederman [Thu, 4 Feb 2010 07:13:24 +0000 (23:13 -0800)]
sysfs: sysfs_sd_setattr set iattrs unconditionally

There is currently a bug in sysfs_sd_setattr inherited from
sysfs_setattr in 2.6.32 where the first time we set the attributes
on a sysfs file we allocate backing store but do not set the
backing store attributes.  Resulting in overly restrictive
permissions on sysfs files.

The fix is to simply modify the code so that it always executes
when we update the sysfs attributes, as we did in 2.6.31 and earlier.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Tested-by: Jean Delvare <khali@linux-fr.org>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agosparc64: Kill bogus ip_tables.h include.
David S. Miller [Tue, 16 Feb 2010 23:23:23 +0000 (15:23 -0800)]
sparc64: Kill bogus ip_tables.h include.

Fixes the following build failure:

  CC      arch/sparc/kernel/sys_sparc32.o
In file included from include/linux/netfilter_ipv4/ip_tables.h:28,
                 from arch/sparc/kernel/sys_sparc32.c:46:
include/linux/netfilter/x_tables.h:525: error: expected declaration specifiers or â€˜...’ before â€˜nf_hookfn’

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Update firmwares and update version to 2.0.8.
Michael Chan [Tue, 16 Feb 2010 23:19:42 +0000 (15:19 -0800)]
bnx2: Update firmwares and update version to 2.0.8.

- Increase FTQ depth to 256 to ehnabce performance.
- Fix RV2P context corruption on 5709 when flow control is enabled.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Fix bug when saving statistics.
Patrick Rabau [Mon, 15 Feb 2010 19:42:11 +0000 (19:42 +0000)]
bnx2: Fix bug when saving statistics.

This fixes the problem of dropping the carry when adding 2 32-bit values.
Switch to use array indexing for better readability.

Reported by and fix provided by Patrick Rabau.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Allow user-specified multiple advertisement speed values.
Michael Chan [Mon, 15 Feb 2010 19:42:10 +0000 (19:42 +0000)]
bnx2: Allow user-specified multiple advertisement speed values.

Remove unnecessary code that works around older versions of ethtool
that can pass down invalid advertisement speed values.  This old
code prevents the user from specifying multiple advertisement values.
The new code uses simple masking to mask out invalid advertisment bits.

Reported-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Adjust flow control water marks.
Michael Chan [Mon, 15 Feb 2010 19:42:09 +0000 (19:42 +0000)]
bnx2: Adjust flow control water marks.

The current water marks are too high and can cause unnecessary flow
control frames.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Need to call cnic_setup_cnic_irq_info() after MTU change.
Michael Chan [Mon, 15 Feb 2010 19:42:08 +0000 (19:42 +0000)]
bnx2: Need to call cnic_setup_cnic_irq_info() after MTU change.

New status blocks are allocated during MTU change so we need to
update this information for the cnic driver.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Check BNX2_FLAG_USING_MSIX flag when setting up MSIX.
Michael Chan [Mon, 15 Feb 2010 19:42:07 +0000 (19:42 +0000)]
bnx2: Check BNX2_FLAG_USING_MSIX flag when setting up MSIX.

Checking the flag is more correct than checking bp->irq_nvecs. By
accident it is not a problem because we always have more than 1
vectors when using MSIX mode.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoatl1c: Add support for Atheros AR8152 and AR8152
Luis R. Rodriguez [Tue, 16 Feb 2010 23:16:45 +0000 (15:16 -0800)]
atl1c: Add support for Atheros AR8152 and AR8152

AR8151 is a Gigabit Ethernet device. AR8152 devices are
Fast Ethernet devices, there are two revisions, a 1.0
and a 2.0 revision.

This has been tested against these devices:

Driver Model-name vendor:device Type
atl1c  AR8131 1969:1063 Gigabit Ethernet
atl1c AR8132 1969:1062 Fast Ethernet
atl1c AR8151(v1.0) 1969:1073 Gigabit Ethernet
atl1c AR8152(v1.1) 1969:2060 Fast Ethernet

This device has no hardware available yet so it goes untested,
but it should work:

atl1c AR8152(v2.0) 1969:2062 Fast Ethernet

Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoUSB: gadget: fix EEM gadget CRC usage
Brian Niebuhr [Mon, 25 Jan 2010 20:45:40 +0000 (14:45 -0600)]
USB: gadget: fix EEM gadget CRC usage

eem_wrap() is sending a sentinel CRC, but it didn't indicate that to
the host, it should zero bit 14 (bmCRC) in the EEM packet header,
instead of setting it.

Also remove a redundant crc calculation in eem_unwrap().

Signed-off-by: Steve Longerbeam <stevel@netspectrum.com>
Acked-by: Brian Niebuhr <bniebuhr@efjohnson.com>
Acked-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: otg Kconfig: let USB_OTG_UTILS select USB_ULPI option
Valentin Longchamp [Wed, 20 Jan 2010 19:06:31 +0000 (20:06 +0100)]
USB: otg Kconfig: let USB_OTG_UTILS select USB_ULPI option

With CONFIG_USB_ULPI=y, CONFIG_USB<=m, CONFIG_PCI=n and
CONFIG_USB_OTG_UTILS=n, which is the default used for mx31moboard,
the build for all mx3 platforms fails because drivers/usb/otg/ulpi.c
where otg_ulpi_create is defined is not compiled.

Build error:
arch/arm/mach-mx3/built-in.o: In function `mxc_board_init':
kzmarm11.c:(.init.text+0x73c): undefined reference to `otg_ulpi_create'
kzmarm11.c:(.init.text+0x1020): undefined reference to `otg_ulpi_create'

This isn't a strong dependency as drivers/usb/otg/ulpi.c doesn't
use functions defined in drivers/usb/otg/otg.o and is only needed
to get ulpi.o linked into the kernel image.

Signed-off-by: Valentin Longchamp <valentin.longchamp@epfl.ch>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: g_multi: fix CONFIG_USB_G_MULTI_RNDIS usage
Michal Nazarewicz [Fri, 22 Jan 2010 14:18:21 +0000 (15:18 +0100)]
USB: g_multi: fix CONFIG_USB_G_MULTI_RNDIS usage

g_multi used CONFIG_USB_ETH_RNDIS to check if RNDIS option was requested
where it should check for CONFIG_USB_G_MULTI_RNDIS.  As a result, RNDIS
was never present in g_multi regardless of configuration.

This fixes changes made in commit 396cda90d228d0851f3d64c7c85a1ecf6b8ae1e8.

Signed-off-by: Michal Nazarewicz <m.nazarewicz@samsung.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agokfifo: Don't use integer as NULL pointer
Anton Vorontsov [Wed, 27 Jan 2010 14:09:38 +0000 (17:09 +0300)]
kfifo: Don't use integer as NULL pointer

This patch fixes following sparse warnings:

include/linux/kfifo.h:127:25: warning: Using plain integer as NULL pointer
kernel/kfifo.c:83:21: warning: Using plain integer as NULL pointer

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: Stefani Seibold <stefani@seibold.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: FHCI: Fix build after kfifo rework
Anton Vorontsov [Wed, 27 Jan 2010 14:09:36 +0000 (17:09 +0300)]
USB: FHCI: Fix build after kfifo rework

After kfifo rework FHCI fails to build:

  CC      drivers/usb/host/fhci-tds.o
drivers/usb/host/fhci-tds.c: In function 'fhci_ep0_free':
drivers/usb/host/fhci-tds.c:108: error: used struct type value where scalar is required
drivers/usb/host/fhci-tds.c:118: error: used struct type value where scalar is required
drivers/usb/host/fhci-tds.c:128: error: used struct type value where scalar is required

This is because kfifos are no longer pointers in the ep struct.
So, instead of checking the pointers, we should now check if kfifo
is initialized.

Reported-by: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: Stefani Seibold <stefani@seibold.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agokfifo: Make kfifo_initialized work after kfifo_free
Anton Vorontsov [Wed, 27 Jan 2010 14:09:34 +0000 (17:09 +0300)]
kfifo: Make kfifo_initialized work after kfifo_free

After kfifo rework it's no longer possible to reliably know if kfifo is
usable, since after kfifo_free(), kfifo_initialized() would still return
true. The correct behaviour is needed for at least FHCI USB driver.

This patch fixes the issue by resetting the kfifo to zero values (the
same approach is used in kfifo_alloc() if allocation failed).

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: Stefani Seibold <stefani@seibold.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: serial: add usbid for dell wwan card to sierra.c
Richard Farina [Wed, 20 Jan 2010 21:42:33 +0000 (16:42 -0500)]
USB: serial: add usbid for dell wwan card to sierra.c

This patch adds support for Dell Computer Corp. Wireless 5720 VZW Mobile
Broadband (EVDO Rev-A) Minicard GPS Port.  I stole the name from lsusb,
but my card does not have a GPS on it (at least not that I can make
function). I'm sure the patch is whitespace damaged but the one line
addition should be fairly straightforward nonetheless.

Tested-by: Rick Farina <sidhayn@gmail.com>
Signed-off-by: Rick Farina <sidhayn@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: SIS USB2VGA DRIVER: support KAIREN's USB VGA adaptor USB20SVGA-MB-PLUS
Tanaka Akira [Wed, 20 Jan 2010 17:31:09 +0000 (02:31 +0900)]
USB: SIS USB2VGA DRIVER: support KAIREN's USB VGA adaptor USB20SVGA-MB-PLUS

This patch adds the USB product ID of KAIREN's USB VGA Adaptor,
USB20SVGA-MB-PLUS, to sisusbvga work with it.

Signed-off-by: Tanaka Akira <akr@fsij.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: ehci: phy low power mode bug fixing
Alek Du [Tue, 19 Jan 2010 08:31:31 +0000 (16:31 +0800)]
USB: ehci: phy low power mode bug fixing

1. There are two msleep calls inside two spin lock sections, need to unlock
   and lock again after msleep.
2. Save a extra status reg setting.

Signed-off-by: Alek Du <alek.du@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: s3c-hsotg: Export usb_gadget_register_driver()
Mark Brown [Mon, 18 Jan 2010 13:18:34 +0000 (13:18 +0000)]
USB: s3c-hsotg: Export usb_gadget_register_driver()

USB gadget controller drivers normally export their driver registration
function, allowing modular builds of the individual gadget drivers so
do so for s3c-hsotg, fixing builds.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: r8a66597-udc: Prototype IS_ERR() and PTR_ERR()
Mark Brown [Mon, 18 Jan 2010 12:03:18 +0000 (12:03 +0000)]
USB: r8a66597-udc: Prototype IS_ERR() and PTR_ERR()

The build of r8a66597-udc was failing on ARM since IS_ERR() and
PTR_ERR() weren't protyped.  Presumably err.h is being pulled in by
another header on other platforms.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Yoshihiro Shimoda <shimoda.yoshihiro@renesas.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: ftdi_sio: add device IDs (several ELV, one Mindstorms NXT)
Andreas Mohr [Sun, 17 Jan 2010 10:45:38 +0000 (11:45 +0100)]
USB: ftdi_sio: add device IDs (several ELV, one Mindstorms NXT)

- add FTDI device IDs for several ELV devices and NXTCam of Lego Mindstorms NXT
- add hopefully helpful new_id comment
- remove less helpful "Due to many user requests for multiple ELV devices we enable
  them by default." comment (we simply add _all_ known devices - an
  enduser shouldn't have to fiddle with obscure module parameters...).
- add myself to DRIVER_AUTHOR

The missing NXTCam ID has been found at
http://www.unixboard.de/vb3/showthread.php?t=44155
, ELV devices taken from ELV Windows .inf file.

Signed-off-by: Andreas Mohr <andi@lisas.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: storage: Remove unneeded SC/PR from unusual_devs.h
Phil Dibowitz [Sat, 16 Jan 2010 18:52:17 +0000 (19:52 +0100)]
USB: storage: Remove unneeded SC/PR from unusual_devs.h

This patch removes the subclass and protocol entries from a Microtech
entry in unusual_devs.h. This was reported by <ryck@pacbell.net>.

Greg, please apply.

Signed-off-by: Phil Dibowitz <phil@ipom.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: ftdi_sio: new device id for papouch AD4USB
Radek Liboska [Wed, 27 Jan 2010 14:38:34 +0000 (15:38 +0100)]
USB: ftdi_sio: new device id for papouch AD4USB

added new device pid (PAPOUCH_AD4USB_PID) to ftdi_sio.h and ftdi_sio.c

AD4USB measuring converter is a 4-input A/D converter which enables the
user to measure to four current inputs ranging from 0(4) to 20 mA or
voltage between 0 and 10 V. The measured values are then transferred to
a superior system in digital form. The AD4USB communicates via USB.
Powered is also via USB.  datasheet in english is here:
http://www.papouch.com/shop/scripts/pdf/ad4usb_en.pdf

Signed-off-by: Radek Liboska <liboska@uochb.cas.cz>
14 years agoUSB: usbfs: properly clean up the as structure on error paths
Linus Torvalds [Tue, 16 Feb 2010 20:35:07 +0000 (12:35 -0800)]
USB: usbfs: properly clean up the as structure on error paths

I notice that the processcompl_compat() function seems to be leaking the
'struct async *as' in the error paths.

I think that the calling convention is fundamentally buggered. The
caller is the one that did the "reap_as()" to get the as thing, the
caller should be the one to free it too.

Freeing it in the caller also means that it very clearly always gets
freed, and avoids the need for any "free in the error case too".

From: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Marcus Meissner <meissner@suse.de>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoUSB: usbfs: only copy the actual data received
Greg KH [Mon, 15 Feb 2010 17:37:46 +0000 (09:37 -0800)]
USB: usbfs: only copy the actual data received

We need to only copy the data received by the device to userspace, not
the whole kernel buffer, which can contain "stale" data.

Thanks to Marcus Meissner for pointing this out and testing the fix.

Reported-by: Marcus Meissner <meissner@suse.de>
Tested-by: Marcus Meissner <meissner@suse.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agotunnels: fix netns vs proto registration ordering
Alexey Dobriyan [Tue, 16 Feb 2010 09:05:04 +0000 (09:05 +0000)]
tunnels: fix netns vs proto registration ordering

Same stuff as in ip_gre patch: receive hook can be called before netns
setup is done, oopsing in net_generic().

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agogre: fix netns vs proto registration ordering
Alexey Dobriyan [Tue, 16 Feb 2010 07:57:44 +0000 (07:57 +0000)]
gre: fix netns vs proto registration ordering

GRE protocol receive hook can be called right after protocol addition is done.
If netns stuff is not yet initialized, we're going to oops in
net_generic().

This is remotely oopsable if ip_gre is compiled as module and packet
comes at unfortunate moment of module loading.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobe2net: set proper value to version field in req hdr
Ajit Khaparde [Tue, 16 Feb 2010 00:18:46 +0000 (00:18 +0000)]
be2net: set proper value to version field in req hdr

Before sending a command to the ASIC, set version properly.
This is necessary for the ARM firmware to send correct data to the driver.
This also fixes a bug in certain skews of the ASIC where the statistics
are misreported.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoxfrm: Fix xfrm_state_clone leak
Herbert Xu [Mon, 15 Feb 2010 20:00:51 +0000 (20:00 +0000)]
xfrm: Fix xfrm_state_clone leak

xfrm_state_clone calls kfree instead of xfrm_state_put to free
a failed state.  Depending on the state of the failed state, it
can cause leaks to things like module references.

All states should be freed by xfrm_state_put past the point of
xfrm_init_state.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoipcomp: Avoid duplicate calls to ipcomp_destroy
Herbert Xu [Mon, 15 Feb 2010 19:24:30 +0000 (19:24 +0000)]
ipcomp: Avoid duplicate calls to ipcomp_destroy

When ipcomp_tunnel_attach fails we will call ipcomp_destroy twice.
This may lead to double-frees on certain structures.

As there is no reason to explicitly call ipcomp_destroy, this patch
removes it from ipcomp*.c and lets the standard xfrm_state destruction
take place.

This is based on the discovery and patch by Alexey Dobriyan.

Tested-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoethtool: allow non-admin user to read GRO settings.
stephen hemminger [Thu, 11 Feb 2010 13:48:29 +0000 (13:48 +0000)]
ethtool: allow non-admin user to read GRO settings.

Looks like an oversight in GRO design.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-2.6-dm
Linus Torvalds [Tue, 16 Feb 2010 20:22:15 +0000 (12:22 -0800)]
Merge git://git./linux/kernel/git/agk/linux-2.6-dm

* git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-2.6-dm:
  dm: sysfs revert add empty release function to avoid debug warning
  dm mpath: fix stall when requeueing io
  dm raid1: fix null pointer dereference in suspend
  dm raid1: fail writes if errors are not handled and log fails
  dm log: userspace fix overhead_size calcuations
  dm snapshot: persistent annotate work_queue as on stack
  dm stripe: avoid divide by zero with invalid stripe count

14 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Tue, 16 Feb 2010 19:59:01 +0000 (11:59 -0800)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] preserve personality flag bits across exec

14 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/kaber/nf-next-2.6
David S. Miller [Tue, 16 Feb 2010 19:15:13 +0000 (11:15 -0800)]
Merge branch 'master' of git://git./linux/kernel/git/kaber/nf-next-2.6

14 years agodm: sysfs revert add empty release function to avoid debug warning
Alasdair G Kergon [Tue, 16 Feb 2010 18:43:04 +0000 (18:43 +0000)]
dm: sysfs revert add empty release function to avoid debug warning

Revert commit d2bb7df8cac647b92f51fb84ae735771e7adbfa7 at Greg's request.

    Author: Milan Broz <mbroz@redhat.com>
    Date:   Thu Dec 10 23:51:53 2009 +0000

    dm: sysfs add empty release function to avoid debug warning

    This patch just removes an unnecessary warning:
     kobject: 'dm': does not have a release() function,
     it is broken and must be fixed.

    The kobject is embedded in mapped device struct, so
    code does not need to release memory explicitly here.

Cc: Greg KH <gregkh@suse.de>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
14 years agodm mpath: fix stall when requeueing io
Kiyoshi Ueda [Tue, 16 Feb 2010 18:43:01 +0000 (18:43 +0000)]
dm mpath: fix stall when requeueing io

This patch fixes the problem that system may stall if target's ->map_rq
returns DM_MAPIO_REQUEUE in map_request().
E.g. stall happens on 1 CPU box when a dm-mpath device with queue_if_no_path
     bounces between all-paths-down and paths-up on I/O load.

When target's ->map_rq returns DM_MAPIO_REQUEUE, map_request() requeues
the request and returns to dm_request_fn().  Then, dm_request_fn()
doesn't exit the I/O dispatching loop and continues processing
the requeued request again.
This map and requeue loop can be done with interrupt disabled,
so 1 CPU system can be stalled if this situation happens.

For example, commands below can stall my 1 CPU box within 1 minute or so:
  # dmsetup table mp
  mp: 0 2097152 multipath 1 queue_if_no_path 0 1 1 service-time 0 1 2 8:144 1 1
  # while true; do dd if=/dev/mapper/mp of=/dev/null bs=1M count=100; done &
  # while true; do \
  > dmsetup message mp 0 "fail_path 8:144" \
  > dmsetup suspend --noflush mp \
  > dmsetup resume mp \
  > dmsetup message mp 0 "reinstate_path 8:144" \
  > done

To fix the problem above, this patch changes dm_request_fn() to exit
the I/O dispatching loop once if a request is requeued in map_request().

Signed-off-by: Kiyoshi Ueda <k-ueda@ct.jp.nec.com>
Signed-off-by: Jun'ichi Nomura <j-nomura@ce.jp.nec.com>
Cc: stable@kernel.org
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
14 years agodm raid1: fix null pointer dereference in suspend
Takahiro Yasui [Tue, 16 Feb 2010 18:42:58 +0000 (18:42 +0000)]
dm raid1: fix null pointer dereference in suspend

When suspending a failed mirror, bios are completed by mirror_end_io() and
__rh_lookup() in dm_rh_dec() returns NULL where a non-NULL return value is
required by design.  Fix this by not changing the state of the recovery failed
region from DM_RH_RECOVERING to DM_RH_NOSYNC in dm_rh_recovery_end().

Issue

On 2.6.33-rc1 kernel, I hit the bug when I suspended the failed
mirror by dmsetup command.

BUG: unable to handle kernel NULL pointer dereference at 00000020
IP: [<f94f38e2>] dm_rh_dec+0x35/0xa1 [dm_region_hash]
...
EIP: 0060:[<f94f38e2>] EFLAGS: 00010046 CPU: 0
EIP is at dm_rh_dec+0x35/0xa1 [dm_region_hash]
EAX: 00000286 EBX: 00000000 ECX: 00000286 EDX: 00000000
ESI: eff79eac EDI: eff79e80 EBP: f6915cd4 ESP: f6915cc4
 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
Process dmsetup (pid: 2849, ti=f6914000 task=eff03e80 task.ti=f6914000)
 ...
Call Trace:
 [<f9530af6>] ? mirror_end_io+0x53/0x1b1 [dm_mirror]
 [<f9413104>] ? clone_endio+0x4d/0xa2 [dm_mod]
 [<f9530aa3>] ? mirror_end_io+0x0/0x1b1 [dm_mirror]
 [<f94130b7>] ? clone_endio+0x0/0xa2 [dm_mod]
 [<c02d6bcb>] ? bio_endio+0x28/0x2b
 [<f952f303>] ? hold_bio+0x2d/0x62 [dm_mirror]
 [<f952f942>] ? mirror_presuspend+0xeb/0xf7 [dm_mirror]
 [<c02aa3e2>] ? vmap_page_range+0xb/0xd
 [<f9414c8d>] ? suspend_targets+0x2d/0x3b [dm_mod]
 [<f9414ca9>] ? dm_table_presuspend_targets+0xe/0x10 [dm_mod]
 [<f941456f>] ? dm_suspend+0x4d/0x150 [dm_mod]
 [<f941767d>] ? dev_suspend+0x55/0x18a [dm_mod]
 [<c0343762>] ? _copy_from_user+0x42/0x56
 [<f9417fb0>] ? dm_ctl_ioctl+0x22c/0x281 [dm_mod]
 [<f9417628>] ? dev_suspend+0x0/0x18a [dm_mod]
 [<f9417d84>] ? dm_ctl_ioctl+0x0/0x281 [dm_mod]
 [<c02c3c4b>] ? vfs_ioctl+0x22/0x85
 [<c02c422c>] ? do_vfs_ioctl+0x4cb/0x516
 [<c02c42b7>] ? sys_ioctl+0x40/0x5a
 [<c0202858>] ? sysenter_do_call+0x12/0x28

Analysis

When recovery process of a region failed, dm_rh_recovery_end() function
changes the state of the region from RM_RH_RECOVERING to DM_RH_NOSYNC.
When recovery_complete() is executed between dm_rh_update_states() and
dm_writes() in do_mirror(), bios are processed with the region state,
DM_RH_NOSYNC. However, the region data is freed without checking its
pending count when dm_rh_update_states() is called next time.

When bios are finished by mirror_end_io(), __rh_lookup() in dm_rh_dec()
returns NULL even though a valid return value are expected.

Solution

Remove the state change of the recovery failed region from DM_RH_RECOVERING
to DM_RH_NOSYNC in dm_rh_recovery_end(). We can remove the state change
because:

  - If the region data has been released by dm_rh_update_states(),
    a new region data is created with the state of DM_RH_NOSYNC, and
    bios are processed according to the DM_RH_NOSYNC state.

  - If the region data has not been released by dm_rh_update_states(),
    a state of the region is DM_RH_RECOVERING and bios are put in the
    delayed_bio list.

The flag change from DM_RH_RECOVERING to DM_RH_NOSYNC in dm_rh_recovery_end()
was added in the following commit:
  dm raid1: handle resync failures
  author  Jonathan Brassow <jbrassow@redhat.com>
    Thu, 12 Jul 2007 16:29:04 +0000 (17:29 +0100)
  http://git.kernel.org/linus/f44db678edcc6f4c2779ac43f63f0b9dfa28b724

Signed-off-by: Takahiro Yasui <tyasui@redhat.com>
Reviewed-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
14 years agodm raid1: fail writes if errors are not handled and log fails
Mikulas Patocka [Tue, 16 Feb 2010 18:42:55 +0000 (18:42 +0000)]
dm raid1: fail writes if errors are not handled and log fails

If the mirror log fails when the handle_errors option was not selected
and there is no remaining valid mirror leg, writes return success even
though they weren't actually written to any device.  This patch
completes them with EIO instead.

This code path is taken:
do_writes:
bio_list_merge(&ms->failures, &sync);
do_failures:
if (!get_valid_mirror(ms)) (false)
else if (errors_handled(ms)) (false)
else bio_endio(bio, 0);

The logic in do_failures is based on presuming that the write was already
tried: if it succeeded at least on one leg (without handle_errors) it
is reported as success.

Reference: https://bugzilla.redhat.com/show_bug.cgi?id=555197

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
14 years agodm log: userspace fix overhead_size calcuations
Jonathan Brassow [Tue, 16 Feb 2010 18:42:53 +0000 (18:42 +0000)]
dm log: userspace fix overhead_size calcuations

This patch fixes two bugs that revolve around the miscalculation and
misuse of the variable 'overhead_size'.  'overhead_size' is the size of
the various header structures used during communication.

The first bug is the use of 'sizeof' with the pointer of a structure
instead of the structure itself - resulting in the wrong size being
computed.  This is then used in a check to see if the payload
(data_size) would be to large for the preallocated structure.  Since the
bug produces a smaller value for the overhead, it was possible for the
structure to be breached.  (Although the current users of the code do
not currently send enough data to trigger this bug.)

The second bug is that the 'overhead_size' value is used to compute how
much of the preallocated space should be cleared before populating it
with fresh data.  This should have simply been 'sizeof(struct cn_msg)'
not overhead_size.  The fact that 'overhead_size' was computed
incorrectly made this problem "less bad" - leaving only a pointer's
worth of space at the end uncleared.  Thus, this bug was never producing
a bad result, but still needs to be fixed - especially now that the
value is computed correctly.

Cc: stable@kernel.org
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
14 years agodm snapshot: persistent annotate work_queue as on stack
Mike Snitzer [Tue, 16 Feb 2010 18:42:51 +0000 (18:42 +0000)]
dm snapshot: persistent annotate work_queue as on stack

chunk_io() declares its 'struct mdata_req' on the stack and then
initializes its 'struct work_struct' member.  Annotate the
initialization of this workqueue with INIT_WORK_ON_STACK to suppress a
debugobjects warning seen when CONFIG_DEBUG_OBJECTS_WORK is enabled.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
14 years agodm stripe: avoid divide by zero with invalid stripe count
Nikanth Karthikesan [Tue, 16 Feb 2010 18:42:47 +0000 (18:42 +0000)]
dm stripe: avoid divide by zero with invalid stripe count

If a table containing zero as stripe count is passed into stripe_ctr
the code attempts to divide by zero.

This patch changes DM_TABLE_LOAD to return -EINVAL if the stripe count
is zero.

We now get the following error messages:
  device-mapper: table: 253:0: striped: Invalid stripe count
  device-mapper: ioctl: error adding target to table

Signed-off-by: Nikanth Karthikesan <knikanth@suse.de>
Cc: stable@kernel.org
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
14 years agox86: ELF_PLAT_INIT() shouldn't worry about TIF_IA32
Oleg Nesterov [Tue, 16 Feb 2010 14:24:01 +0000 (15:24 +0100)]
x86: ELF_PLAT_INIT() shouldn't worry about TIF_IA32

The 64-bit version of ELF_PLAT_INIT() clears TIF_IA32, but at this point
it has already been cleared by SET_PERSONALITY == set_personality_64bit.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agox86: set_personality_ia32() misses force_personality32
Oleg Nesterov [Tue, 16 Feb 2010 14:02:13 +0000 (15:02 +0100)]
x86: set_personality_ia32() misses force_personality32

05d43ed8a "x86: get rid of the insane TIF_ABI_PENDING bit" forgot about
force_personality32.  Fix.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agonetfilter: ebtables: mark: add CONFIG_COMPAT support
Florian Westphal [Wed, 27 Jan 2010 13:39:05 +0000 (14:39 +0100)]
netfilter: ebtables: mark: add CONFIG_COMPAT support

Add the required handlers to convert 32 bit
ebtables mark match and match target structs to 64bit layout.

Signed-off-by: Florian Westphal <fwestphal@astaro.com>
14 years agonetfilter: ebt_limit: add CONFIG_COMPAT support
Florian Westphal [Wed, 27 Jan 2010 13:38:32 +0000 (14:38 +0100)]
netfilter: ebt_limit: add CONFIG_COMPAT support

ebt_limit structure is larger on 64 bit systems due
to "long" type used in the (kernel-only) data section.

Setting .compatsize is enough in this case, these values
have no meaning in userspace.

Signed-off-by: Florian Westphal <fwestphal@astaro.com>
14 years agonetfilter: ebtables: try native set/getsockopt handlers, too
Florian Westphal [Sun, 7 Feb 2010 02:19:12 +0000 (03:19 +0100)]
netfilter: ebtables: try native set/getsockopt handlers, too

ebtables can be compiled to perform userspace-side padding of
structures. In that case, all the structures are already in the
'native' format expected by the kernel.

This tries to determine what format the userspace program is
using.

For most set/getsockopts, this can be done by checking
the len argument for sizeof(compat_ebt_replace) and
re-trying the native handler on error.

In case of EBT_SO_GET_ENTRIES, the native handler is tried first,
it will error out early when checking the *len argument
(the compat version has to defer this check until after
 iterating over the kernel data set once, to adjust for all
 the structure size differences).

As this would cause error printks, remove those as well, as
recommended by Bart de Schuymer.

Signed-off-by: Florian Westphal <fw@strlen.de>
14 years agonetfilter: ebtables: add CONFIG_COMPAT support
Florian Westphal [Tue, 5 Jan 2010 15:09:46 +0000 (16:09 +0100)]
netfilter: ebtables: add CONFIG_COMPAT support

Main code for 32 bit userland ebtables binary with 64 bit kernels
support.

Tested on x86_64 kernel only, using 64bit ebtables binary
for output comparision.

At least ebt_mark, m_mark and ebt_limit need CONFIG_COMPAT hooks, too.

remaining problem:

The ebtables userland makefile has:
ifeq ($(shell uname -m),sparc64)
CFLAGS+=-DEBT_MIN_ALIGN=8 -DKERNEL_64_USERSPACE_32
endif

struct ebt_replace, ebt_entry_match etc. then contain userland-side
padding, i.e.  even if we are called from a 32 bit userland, the
structures may already be in the right format.

This problem is addressed in a follow-up patch.

Signed-off-by: Florian Westphal <fwestphal@astaro.com>
14 years agonetfilter: ebtables: split update_counters into two functions
Florian Westphal [Sun, 7 Feb 2010 01:48:47 +0000 (02:48 +0100)]
netfilter: ebtables: split update_counters into two functions

allows to call do_update_counters() from upcoming CONFIG_COMPAT
code instead of copy&pasting the same code.

Signed-off-by: Florian Westphal <fw@strlen.de>